The Influence of Cognitive Functioning and Psychopathic Personality Characteristics on the Offense Supportive Cognitions of Pedophiles

Bradley Allen, Dawn Pflugradt, Gina Olson
Sand Ridge Secure Treatment Center, Mauston, USA

[Sexual Offender Treatment, Volume 8 (2013), Issue 2]

Abstract

Despite a substantial body of research which supports the efficacy of cognitive behavioral treatment for sexual offenders, it is not clear whether offense supportive cognitions are expressions of entrenched belief systems which allow offenders to overcome internal inhibitions or if they arise from post-offending conscious processes to reduce psychological distress. The following study utilized a repertory grid technique to evaluate the offense supportive cognitions of a sample of offenders who were diagnosed with Pedophilic Disorder (N=23). The offenders were either civilly committed or referred for commitment. The sample was comprised of participants from four sub-groups: average intellectual functioning/low psychopathy; below average intellectual functioning and low psychopathy; average intellectual functioning/moderate-high psychopathy; and below average intellectual functioning/moderate-high psychopathy. Results indicate a moderate consensus among the participants endorsing offense supportive cognitions. Additionally, post hoc correlation analyses between the total Sex with Children scores and dependent measures were significant for four of the repertory grid elements. Limitations and implications for treatment are discussed.

Key Words: adult sexual offenders, assessment, child victims, pedophilic disorder


Among the various treatment modalities utilized to reduce recidivism risk for sexual offenders, cognitive behavioral approaches have been empirically demonstrated to be among the most efficacious (Hanson, Bourgon, Helmus, & Hodgson, 2009; Lösel & Schmucker, 2005; Hanson, Gordon, Harris, Marques, Murphy, Quinsey, & Seto, 2002). Despite the substantial body of research which supports cognitive behavioral treatment, the operationalization of cognitive processes as they relate to sexual offending behaviors has been hindered by the lack of an integrated theoretical model (Marshall, Marshall, Serran, & O'Brien, 2011; Ward, Hudson, Johnston, & Marshall, 1997). As a result, it is not clear whether "cognitive distortions" (Abel, Becker, & Cunningham-Rather, 1984) or offense supportive cognitions are expressions of entrenched belief systems which allow offenders to overcome internal inhibitions and commit sexual assaults or if they arise from post-offending conscious processes to reduce psychological distress (Finkelhor, 1984). Moreover, it is also not known if, and to what extent, these processes vary across sexual offenders with differing characteristics. This relationship between individual characteristics and thought processes is an important consideration, however, as articulated in the first principle of cognitive behavior therapy, "Cognitive behavior therapy is based upon an ever-evolving formulation of patients' problems and an individual conceptualization of each patient in cognitive terms" (Becker, 2011, p. 22).

Historically, the primary method utilized to identify individual offense supportive cognitions has been psychometric measures of dysfunctional beliefs and attitudes that discriminate sexual offenders from others (Mann, Webster, Wakeling, & Marshall, 2007). Among the first quantitative measures to evaluate the offense supportive cognitions relating to sexual assaults against children was a cognitive distortions scale developed by Abel, Gore, Holland, Camp, Becker, and Rathner (1989). The Abel et al. (1989) scale was found to load on six factors with acceptable inter-item consistency and test-retest reliability, but had poor discriminant validity (Gannon, Keown, & Rose, 2009). That is, even though child molesters endorsed significantly more items than community controls, it did not discriminate between child molesters and non-child molesting individuals with paraphilias. Moreover, the scale was criticized for being too susceptible to socially desirable responding (Langevin, 1991). Interestingly, as suggested by Mann and colleagues (2007), the Abel et al. (1989) scale appeared to support the hypothesis that offense supportive cognitions maintain offending behaviors, not cause them, as they tend to increase in strength over the course of the sexual assaults.

In response to these limitations, Bumby (1996) developed the MOLEST and RAPE scales derived from clinical and research experience. The original 44 item MOLEST scale, which included some items from the Abel et al. (1989) scale, was administered to 44 incarcerated males convicted of sexual assault or attempted sexual assault of a child (Bumby, 1996). The subsequent 38-item final version of the scale was found to have excellent internal consistency with acceptable test-retest reliability, excellent construct and convergent validity and unlike the Abel et al. (1989) scale, no social desirability response bias. Despite several improvements, as compared to previous scales, the MOLEST scale also had limitations: 1) some of the items did not appear reasonably related to sexually offending behaviors; and 2) some of the items assumed that the participants were motivated by a deviant sexual preference toward children (Mann et al., 2007).

The subsequent approaches to operationalize offense supportive cognitions were influenced by the research of Ward and colleagues (2000) who argued that sexual offenders' cognitive distortions emerge from underlying implicit or causal theories about the nature of their victims rather than maladaptive belief systems and attitudes. Ward further suggested that these theories were similar to scientific theories insofar as they are used to explain empirical regularities and to make predictions about the world. Using the Sex with Children Scale developed by William L. Marshall and adopted by Her Majesty's Prison Service in the United Kingdom in 1992, Mann et al. (2007) tested Ward's Implicit Theory Model by exploring the relationship between sexual offenders' generalized beliefs and specific cognitions. The Sex with Children Scale is comprised of 18 statements derived from clinical experience which are thought to be reflective of offenders' attempts to justify sex with children. The scale requires participants to rate the statements on a Likert type of metric intended to measure the general beliefs that justify sexual contact between adults and children.

As Mann and colleagues (2007) found, the Sex with Children Scale had excellent reliability, predicative validity, discriminant and concurrent validity and it identified a relationship between the offenders' general offense beliefs and those cognitions specific to their own offense(s). Moreover, a principal component analysis extracted two factors which explained 57% of the common variance. One factor included 11 items labeled "Harmless Sex with Children" (items 8-18) which incorporate beliefs that sexual abuse of children is harmless. In addition, seven items loaded on a second factor, labeled "Provocative Sexual Children" (items 1-7), which relates to beliefs that children are sexual beings who provoke sexual behavior from adults. Interestingly, these factors are similar to two of Ward and Keenan's (1999) hypothesized implicit theories of child molesters: "Children as Sexual Objects" (children desire and enjoy sex); and "Nature of Harm" (sex is always beneficial and therefore is unlikely to harm children).

These findings are consistent with previous research which found that an important consideration when providing treatment for sexual offenders is the underlying offense supportive beliefs; otherwise, challenging their offense specific cognitions will not necessarily generalize to future potential victims (Hanson, Gizzarelli, & Scott, 1994). This distinction was articulated by Blumenthal, Gudjonsson, and Burns (1999) who distinguished offense-specific cognitions (i.e., post hoc justifications and minimizations) from enduring, situationally non-specific cognitions, which focus on the general acceptability of the behavior. The latter conceptualization, also referred to as offense-supported attitudes and beliefs (Maruna & Mann, 2006), had a small but consistent relationship with sexual recidivism (Helmus, Hanson, Babchisin, & Mann, 2013).

Whereas some studies have considered the effect of various psychological characteristics on cognitive distortions (Gannon & Alleyne, 2013; Strickland, 2008; Wegan, Kavanagh, & Blair, 2005; Ward, Hudson, Johnston, & Marshall, 1997), there is no empirical research which has evaluated if offense supportive cognitions are associated with specific psychological characteristics such as cognitive abilities and personality traits. Moreover, although it has been recognized that the efficacy of specific treatment approaches vary across groups with common characteristics such as an intellectually disability (Lambrick & Glaser, 2004; Broxhome & Lindsay, 2003) and psychopathy (Doren, 2007; Harkins, Beech, & Thornton, 2013), the psychological processes eliciting these differences have not been comprehensively researched. By identifying how intellectual functioning and personality characteristics relate to the thought processes of sexual offenders, it may be possible to design more individualized, and thus more effective, treatment approaches.

The following exploratory study examined the association between psychological characteristics (i.e., psychopathy and full-scale I.Q. ) and offense supportive cognitions based upon a personal construct paradigm developed by Kelly (1955). As described by Blagden, Winder, Gregson, and Thorne (2012), personal construct psychology acknowledges that people construe things differently or idiosyncratically, in accordance with their life experiences. That is, people come to understand the world by developing a personally organized system of interpretations or constructs of experiences which are essentially abstractions of a repeated event or set of events (Beail, 1985). Additionally, these psychological processes are adapted or channelled by the ways in which she/he anticipates events (Kelly, 2003). Thus, a person actively construes or interprets events which she/he experiences or anticipates through a structured network of pathways (i.e., personal construct system) (Houston, 1998) which gives it meaning (Kelly, 1991). In short, a person's view of the world is the result of transparent patterns or templates that she/he constructs and applies to her/his personal experiences and realities that makes sense to her/him (Blagden et al., 2012).

Whereas a wide variety of assessment techniques have been developed from personal construct psychology, the vast majority utilize repertory grids (Walker & Winter, 2007). In general, a repertory grid is a set of idiopathic techniques that offers insight into how an individual constructs or construes aspects of her or his reality. As delineated by Jankowicz (2004), each repertory grid is constructed about a particular area of interest or "topic." The topic is represented by "elements" which are usually people but they can be based upon other characteristics of a topic such as places, social roles, etc. "Constructs," the way that a person makes sense of something or derives meaning from it, are then elicited from the elements (e.g., rude-polite, warm-cold, happy-sad, etc.). Each construct is rated within each element, usually with a Likert type of scale, to provide a description of how the person interprets that topic.

A repertory grid technique for studying personal and interpersonal systems of meaning offers many advantages including: 1) providing both quantitative and qualitative information (i.e., endorsing or identifying constructs from various perspectives, respectively); 2) minimal response bias, because it is difficult to see through the aims of the grid, thus making false responses more difficult (i.e., acknowledging types of offense supportive cognitions from neutral, non-threatening, perspectives); 3) allowing for more fundamental positions to be uncovered, due to the indirectness of the method; and 4) allowing for the examination of the values that underlie a person's process of construing his world (Blagden, et al., 2012). Perhaps, the most significant strength of such an approach is that it allows a multi-dimensional evaluation of offense supportive cognitions. That is, this analytical technique provides an opportunity to simultaneously evaluate the participants' offense specific cognitions via constructs and their situationally non-specific cognitions by examining elements.

Although a strength of repertory grid analysis is a reduction of response bias, a major limitation of self-report research (Blagden, et al., 2012), is that it is still susceptible to hindsight bias. As described by Roese and Vohs (2012), "hindsight bias" comprises any combination of three aspects: 1) memory distortion; 2) beliefs about the objective likelihood of an event; and 3) subjective beliefs about one's own prediction abilities. These areas are typically expressed by individuals as selectively recalling information consistent with what they now know to be true, imposing meaning on that information, misattributing an outcome to its assumed prior likelihood, needing to see the world as orderly or predictable and avoiding blame for problems. The consequences of hindsight bias may include an over attention to a single causal understanding of the past and a general overconfidence in the certainty of one's judgments (Roese & Vohs, 2012).

With these strengths and limitations in mind, a repertory grid technique was utilized to evaluate the offense supportive cognitions of a sample of civilly committed (or referred for civil commitment) male sexual offenders from four different sub-groups to ensure that the influence of specific traits/characteristics (i.e., psychopathy and intellectual functioning) was equally represented in the group analysis. As noted, one group had one less participant and consequently, the data were "centered" to minimize any resultant analytical biases. The sub- groups included: average intelligence with low psychopathy (AI/LP); average intelligence with moderate to high psychopathy (AI/HP); below average intelligence with low psychopathy BI/LP); and below average intelligence with moderate to high psychopathy (BI/HP). The primary goal was to evaluate if the levels of psychopathic personality characteristics and levels of cognitive functioning were associated with the frequency and types of offense supportive cognitions expressed by pedophiles. Based upon the assumptions that pedophiles construe their sexually offending behaviors idiosyncratically in accordance with their life experiences through a personal construct system, the primary study hypotheses included:

• Pedophiles will construe or interpret offense supportive cognitions similarly as evidenced by their consensus between elements when controlling for psychopathic personality traits and intellectual functioning.
• The offense supportive cognitions endorsed by pedophiles within each element will be similar when controlling for psychopathic personality traits and intellectual functioning.

Method

Participants

The participants of this study were 23 civilly committed (or referrd for commitment) adult male sexual offenders confined to a secure treatment facility. The primary inclusion critieria were: 1) the participant had a diagnosis of pedophilic disorder including a documented history of sexually assaulting a prepubescent child or children; 2) the participant volunteered to participate in the study; 3) he acknowledged informed consent both in writing and orally; and 4) he had adequate reading skills to complete the study forms and test protocols. The participants who met the inclusion criteria were assigned to one of four groups based upon their full scale I.Q. as measured by the Wechsler Adult Intelligence Scale, Third Edition (WAIS-III) as well as their psychopathic personality characteristics, as measured by the Hare Psychopathy Checklist-Revised (PCL-R).

Measures

The participants with a full-scale I.Q. of 84 or below as measured by the Wechsler Adult Intelligence Scale, Third Edition,were assigned to one of the two below average intelligence groups. Moreover, the participants' developmental and educational histories were reviewed to ensure that their adaptive functioning was consistent with their cognitive abilities. In two cases, participants with higher full-scale I.Q.s (i.e., upper eighties) were assigned to the below average intelligence group due to significantly below average adaptive functioning. That is, not only did they have a history of adaptive behavior deficits, they also required special assistance during institutional programming, similar to other participants with below average intellectual functioning.

The personality characteristics examined by this study were those delineated by Hare (2003) in the Psychopathy Checklist-Revised (PCL-R) which conceptualizes psychopathy as, "a characteristic pattern of interpersonal, affective, and socially deviant behavioral features".Those participants with total PCL-R scores of 26 or greater were assigned to the high psychopathy groups. The cut-off score was determined based upon research that sexualoffenders with total scores of 26 and above, combined with sexual deviancy, are at greater risk to sexually re-offend (Hildebrand, de Ruiter, & de Vogel, 2004; Rice & Harris, 1997).

One participant with a total PCL-R score of 24 was included in a high psychopathy group because his offense history and recent behaviors indicated that it was likely an under estimation of his personality characteristics. Although the original number of participants chosen for the study was 24, six for each group, one person was unable to participate because he was transferred to another faciltiy during the study period. Each participant was provided with a copy of his signed informed consent form and he was encouraged to contact the principal investigator or research assistant if he had further questions or concerns after the interview.

Procedure

After obtaining approval from the facility's Institutional Review Board, annoucements describing the study were placed/posted for two weeks in common areas of the institution. A research assistant/study author reviewed each response and screened the persons who expressed an interest in participating to determine if they met the inclusion criteria. Of the participants who met criteria (i.e., previously diagnosed with pedophilic disorder and whose record contained documented sexual assaults against a prepubescent child[ren]), six were randomly selected from each of the four groups (i.e., AI/LP, AI/HP, BI/LP and BI/HP) and appointments were set for them to meet with the principal investigator. Group assignment was based on file information as well as how they were classified in the secure treatment facility. To ensure confidentiality, the participants were assigned a code number and their identities were known only to the principal investigator. After obtaining both oral and written consent, the interviewer queried the participants using the interview format provided by Mann and their responses were recorded in written notes. The participants were then asked to complete a questionnaire which included the 18 items (i.e., constructs) of the Sex with Children scale from the 13 different persectives (i.e., elements).

Analysis

The data were analysed using repertory grid methodologies derived from personal construct psychology methodologies. As originated by Kelly (1955), the basic grid format is comprised of four components: topic, elements, constructs and ratings (Jankowicz, 2004). Although the processes by which repertory grids are developed have many variations, the general procedure described by Kelly first elicits elements from a topic of interest and then, elicits constructs from the distinctions participants make between the elements.

Other variations of eliciting repertory grids have included methods which provide both the elements and constructs to participants especially when the goal is to examine how a sample of people construe a topic or issue (Bell, 2005; Jankowicz, 2004). Since the focus of this study was comparing the offense supportive cognitions of different groups of pedophiles, the repertory grids constructed for this study were based upon elements and constructs provided by the interviewer. The specific elements used here were adopted from a study by Bladgen and colleagues (2012): Me Now; Me Before Arrest; Me As I'd Like To Be; Father; Mother; Spouse (Wife); Friend; Person You Don't Like; Sexual Offender; Prisoner Admitting Offenses; Prisoner Maintaining Innocence; Police Officer; and Victims. The interviewer also provided the constructs to the participants which were comprised of the 18 statements included in the Sex with Children Scale (see Appendix listing Sex with Children items). Participants were asked to rate the statements (constructs) from each of the role perspectives (elements) using a Likert type scale with the following meter: 0=Strongly Disagree; 1=Disagree; 2=Undecided; 3=Agree; and 4=Strongly Agree.

The resultant data were evaluated with the statistical program Idiogrid (Version 2.4) which is based upon a Personal Construct Psychology paradigm and utilizes an idiographic approach to evaluate personality (Grice, 2002). Whereas most of the analyses included in this program are designed to model data from a single participant, there are also some techniques which allow comparison across individuals. For this study, three Idiogrid analyses were used to evaluate the repertory grids. The first analysis was a Generalized Procrustes Analysis (GPA) which is a multivariate statistical procedure for evaluating multiple grids. It involves the construction of a consensus grid by scaling and rotating the elements or constructs in the individual grids to a point of maximal agreement and then obtaining average results (Grice, 2002).

The "centering" scaling option was used for this analytical procedure which subtracts the grid values for each participant from their respective construct means before submitting them to the GPA algorithm. As described by Grice and Assad (2009), the mean of each centered construct is set to equal zero and potential individual differences in construct means are removed. The consensus grid is saved and used to assess the "consensus view" among the individual grids. The rescaled and rotated individual grids are then compared to the consensus grid by using standard ANOVA formula and procedures (Grice and Assad, 2009). The ANOVA results yield an overall index of variability as "consensus proportion." When converted to percentages, this index ranges from 0-100; the higher values indicate greater similarity across participants. The consensus proportion can be computed as a ratio of the consensus sum of squares and total sum of squares for the ANOVA which is similar to the R2 statistic in regression. Furthermore, this ratio can then be tested for statistical significance using a randomization test.

Secondly, included in the Idiogrid Generalized Procrustes Analysis is an output option "Randomization Test" which is an approach developed by Wakeling, Raats and Halliday (1992) to address the following question: "How much better is the degree of agreement among the grids compared to grids that are comprised of random combinations of constructs or elements" (Idiogrid manual, p.71). Described in different terms, this test analyses the degree of agreement among participant grids to grids that are comprised of random combinations of constructs and elements. This is an important consideration insofar as when randomly generated matrices of numbers are rotated, their level of agreement can be very high (e.g., 50-70%). Essentially, this test provides information so that the user of the GPA may determine if the degree of agreement among the grids is better than what might be obtained from random numbers.

The third Idiogrid procedure was is a Principal Component Analysis (PCA) which is a multivariate statistical procedure for summarizing the numerical information contained in individual repertory grids (Grice, 2002). The purpose of analyzing the consensus grid using a PCA is to reduce its dimensionality so that it can be examined in a bi-plot (Grice & Assad, 2009). For this study, the two principal components of the constructs were used to create the axes. The constructs of the 23 participants were mapped onto a two-dimensional area formed by the two principal components.

Lastly, a series of post hoc Spearman Rank Correlation analyses were conducted to evaluate any associations between the questionnaire total scores and personality and cognitive characteristics. The Spearman Rank Correlation test is a nonparametric measure of correlation which uses ranks based upon the underlying assumptions that the data is a simple random sample measured on at least an ordinal scale (Daniel, 1999). The analyses were conducted for all thirteen elements separately. The first correlation analysis utilized the total scores of the questionnaire and the PCL-R total scores. Next, a correlation analysis was conducted using the questionnaire total scores and full-scale I.Q. scores.

Results

As summarized in Table 1, the mean ages of the participants across the groups were similar ranging from 47 to 49 years. The mean full scale I.Q. of the participants with below average intelligence was 78.3 (SD= 8.6) and the mean I.Q. of participants with average intelligence was 102.4 (SD= 11.4 ). The mean total PCL-R score for participants with low psychopathy was 19.2 (SD=2.4) as compared to the mean total scores of the participants with high psychopathy which was 27 (SD= 1.7). Moreover, the mean Static-99 total scores of the participants were similar across groups ranging from five for three of the groups and six for the average intelligence/moderate-high psychopathy group.

The participants' treatment progress was operationalized by their status in the facility's phase system. The inpatient component of the facility's treatment program has three phases. During phase one, the participant is provided with opportunities to improve his general self-management skills and ameliorate personality disorder symptoms or other general factors that may interfere with his effective treatment participation. In phase two, the participant works with clinical staff to identify the factors that contributed to his offenses and understand their operation in his current functioning. This may include identifying and addressing offense supportive cognitions. Lastly, in phase three the participants learn to manage their re-offense risk factors reliably and replace them with healthy alternative behaviors. The individuals who had not yet started the mainstream treatment program were considered to be in "pre-treatment" status. As illustrated in Table 1, two groups, that is, the AI/LP and BI/HP groups had a greater number of participants in phase one and consequently, they were not yet addressing offense supportive cognitions in treatment. The other two groups, AI/HP and BI/LP, had more participants who were at a stage of treatment (i.e., phase two) that included addressing distorted thoughts associated sexual offending behaviors. Although these participants' treatment level (i.e., progress) may have had some confounding effect, it was likely marginal given that they were in the early stages of identifying offense supportive cognitions.

In regard to other mental health diagnoses besides the inclusion diagnosis of Pedophilic Disorder, the participants' records were also reviewed to determine if they had a major mental illness that may have been symptomatic. That is, if the participant had poor reality testing, it would be difficult (perhaps impossible) to distinguish between delusional ideation from distorted thought processes. As indicated in records, five participants were diagnosed with a Major Depressive Disorder (AI/LP-0; AI/HP-1; BI/LP-3; and BI/HP-1). Additionally, one participant from the average intelligence with moderate to high psychopathy (AI/HP) group was diagnosed with an Unspecified Obsessive Compulsive and Related Disorder. According to available records, none of the participants presented with symptoms or behaviors indicative of a mental illness which would interfere with accurately articulating thoughts and feelings pertaining to their sexual offenses.

The Generalized Procrustes Analysis (GPA) for the personal construct grids resulted in a consensus proportion of 0.44 indicating moderate average agreement among the 23 participants when considering offense supportive cognitions from various perspectives. The consensus proportion was slightly less at 0.32 also indicating moderate agreement between the participants across the elements or perspectives (See Table II). These consensus proportions were tested for statistical significance by using the randomization test described above. The consensus grids for the constructs and elements were analyzed separately; 100 trials were generated for each and not a single value equaled or exceeded the observed consensus proportion. The observed proportions were thus determined to be statistically significant at the p<.01 level.

The ANOVA residuals for the elements and constructs were also computed and examined for points of departure from the consensus grid or matrix. As illustrated in Table III, the residuals ranged in values from 2.93 to 5.39. The participants exhibited the greatest discrepancies in regard to their ratings of the construct "Having sex with a child is not really all that bad because it doesn't really harm the child" (5.39) and "Children are old enough to decide whether or not they want to have sex with someone" ( 4.46). In regard to the elements or perspectives, Table 4 illustrated (residual range 1.45-10.43) that the participants exhibited the greatest amount of discrepancy in the perspectives "Sexual Offender" (10.43) and "Prisoner Maintaining Innocence" (8.06).

A Principal Component Analysis (PCA) chosen for the GPA with a Varimax rotation indicated that two components explained 53.87% of the variance (see Table 5). As depicted in Figure 1, the first component included the constructs: 3. Many children are sexually seductive towards adults; 5. Men who have sex with children are usually led into it by the child; 7. Children who are unloved by their parents are actually helped by men who have sex with them; 8. Sex between adults and children is quite natural and healthy and it is only because of the repressive rules of our society that men are punished for doing this; 9. It is far better for young people to have their first sexual experience during childhood with an adult than to risk what is sure to be an unpleasant experience with someone their own age when they are a teenager. 10. A man can't help having sex with a child if the child acts in a provocative manner; 11. Having sex with a child is not really all that bad because it doesn't really harm the child; 12. If an adult has sex with a child who enjoys it and seems to want it, it shouldn't be considered a crime; 13. Children are old enough to decide whether or not they want to have sex with someone; and 17. It is OK to have sex with a child as long as you don't force the child into it.

The second component included the constructs: 1. Children actually enjoy sex with a an if the man is nice to them; 2. Most children actually enjoy sex with an adult so long as the man does not hurt them; 4. Children enjoy sexual attention from adults; 6. Children who do not wear underwear and who sit in a way that is revealing are suggesting sex; 14. Having sex with a child is a way of expressing your love and affection for that child; 15. Having sex with a child is a good way to teach them about sexuality; 16. Many children benefit from having sex with an adult; and 18. Nowadays it is not so bad to have a sexual relationship with someone who is under-age because kids know so much more about sex than they used to.

Figure 1: Principal Component Analysis (Varimax) of the Consensus Grid

A post hoc nonparametric correlation analysis resulted in a statistically significant negative association between the total scores of the Sex with Children Questionnaire and full scale I.Q. from the perspective of Police Officer (R= -0.46; p=.02). There were also statistically significant negative associations between the total scores of the Sex with Children Questionnaire and PCL-R scores from the perspectives of "Me Before Arrest"( R= -0.46; p=.02), "Sexual Offender" (R= -0.48; p=.01) and "Prisoner Admitting Innocence" (R= -0.53; p<.01).

Discussion

Whereas many treatment approaches for sexual offenders utilize techniques to reduce offense supportive cognitions or thinking (Marshall et al. 2011), there is often definitional and conceptual ambiguity regarding the nature of these thoughts. As suggested by some research, it is more accurate to describe offense supportive cognitions as something offenders "do" rather than something they "have" (Auburn & Lea, 2003). This study utilized a methodology developed from a personal construct psychology paradigm, that is, a repertory grid technique, to evaluate if there was evidence (from a small sample) to suggest that personality traits and level of cognitive functioning are associated with the offense supportive cognitions of pedophiles.

As indicated by a General Procrustes Analysis, the participants' personal construct grids yielded moderate agreement among both the constructs and elements. That is, there was moderate consensus among the participants' construct scores (i.e., ratings of offenses supportive cognitions) and elements (i.e., perspectives of self and others). In general, these results suggest that the levels of cognitive functioning and psychopathic personality traits do not appear to significantly influence the offense supportive cognitions expressed by high risk male sexual offenders. It should be noted, however, that an examination of the ANOVA residuals found significant relative discrepancies for individual constructs and elements. For example, the participants exhibited differences in regard to their ratings of the construct "Having sex with a child is not really all that bad because it doesn't really harm the child" (5.39) and "Children are old enough to decide whether or not they want to have sex with someone" (4.46). Additionally, when examining the element residuals, the participants exhibited the greatest discrepancy in the perspectives "Sexual Offender" (10.43) and "Prisoner Maintaining Innocence" (8.06).

A Principal Component Analysis (PCA) with a Varimax rotation indicated that two components explained 53.87% of the common variance. Whereas this amount of variance for two factors is similar to the findings of the study conducted by Mann et.al (2007), the factors did not contain the same items. Thus, the two latent traits of the Sex with Children Questionnaire identified by Mann and colleagues (2007), that is, "Harmless Sex with Children" and "Provocative Sexual Children," were not identified during the present analysis. Aside from the obvious differences in sample size, these results suggest two possible hypotheses: 1) when controlling for cognitive functioning and psychopathic personality characteristics, the underlying latent structure of offense supportive cognitions differ for high risk sexual offenders, as a group, as compared to lower risk offenders; and 2) the offense supportive cognitions are much more diverse for high risk sexual offenders than other risk groups.

A post hoc analysis of total Sex with Children Questionnaire scores, full scale I.Q. and total PCL-R scores indicated negative correlations for four of the elements or perspectives. First, from the perspective of "Police Officer" lower full scale I.Q.s wereassociated with higher total scores on the Sex with Children Questionnaire. Whereas this result may have been an outlier, it could also be related to an interpretive anomaly (e.g., police officers have some special insight or knowledge) or prior negative experiences with persons in authority. If so, this finding may be indicative of entrenched belief systems.

Secondly, from the perspectives of "Me Before Arrest," "Sexual Offender," and"Prisoner Admitting Innocence," higher PCL-R scores were associated with lower totalscores on the Sex with Children Questionnaire. As suggested for the previous findings, these results may related to interpretive difficulties; given that the responses were limited to three areas pertaining directly to the participants' perspectives over time (i.e. before-during- after offense), however, it is possible to consider that they are at least in part related to post-offending conscious processes to reduce psychological distress.

It should be noted when considering these or any interpretations, a major limitation common to this type of research is that the participants may not have been truthful in their responses (Gannon & Polaschek, 2005). Whereas untruthful ratings likely occurred to some extent, it is hoped that the use of the grid technique as well as the study design adequately mediated this bias. That is, the authors expected that by having participants respond to perceived roles or perspectives other than their own, in the present tense, they would feel less threatened and as a result, more truthful in their responses.

Bearing in mind the limitations of this study and the difficulties with measuring and generalizing such a nebulous concept as offense supportive cognitions, these findings provide a basis to consider the associations between the personality traits, intellectual functioning and cognitions of high risk pedophiles. In general, the repertory grid analysis supports the hypotheses that pedophiles construe or interpret offense supportive cognitions similarly and endorse similar offense supportive cognitions when controlling for psychopathic personality traits and intellectual functioning. The results further indicate, however, that within this group, there is also a great deal of variability. Hopefully, the nature and extent of these differences will be explored in greater depth with larger samples and more comprehensive research designs. As research regarding the assessment and treatment of sexual offenders becomes more refined, there will likely become an increasing awareness that each offender is unique. That is, each offender possesses both common and idiosyncratic psychological characteristics (i.e., nomothetic and idiopathic) that contribute to the initiation and perpetuation of their assaultive behaviors. By identifying these characteristics, assessment and treatment may be adapted or individualized so that recidivism risk is better evaluated and more efficacious therapeutic approaches can be used to reduce that risk.

Appendix

Sex With Children Scale Items

• Children actually enjoy sex with a man if the man is nice to them.
• Most children actually enjoy sex with an adult so long as the man does not hurt them.
• Many children are sexually seductive towards adults.
• Children enjoy sexual attention from adults.
• Men who have sex with children are usually led into it by the child.
• Children who do not wear underwear and who sit in a way that is revealing are suggesting sex.
• Children who are unloved by their parents are actually helped by men who have sex with them.
• Sex between adults and children is quite natural and healthy and it is only because of the repressive rules of our society that men are punished for doing this.
• It is far better for young people to have their first sexual experience during childhood with an adult than to risk what is sure to be an unpleasant experience with someone their own age when they are a teenager.
• A man can't help having sex with a child if the child acts in a provocative manner.
• Having sex with a child is not really all that bad because it doesn't really harm the child.
• If an adult has sex with a child who enjoys it and seems to want it, it shouldn't be considered a crime.
• Children are old enough to decide whether or not they want to have sex with someone.
• Having sex with a child is a way of expressing your love and affection for that child.
• Having sex with a child is a good way to teach them about sexuality.
• Many children benefit from having sex with an adult.
• It is OK to have sex with a child as long as you don't force the child into it.
• Nowadays it is not so bad to have a sexual relationship with someone who is under-age because kids know so much more about sex than they used to.

References

1. Abel, G.G., Gore, D.K., Holland, C.L., Camp, N., Becker, J.V., & Rathner, B.A. (1989). The measurement of the cognitive distortions of child molesters. Annals of Sex Research, 2, 135-153.
2. Abel, G.G., Becker, J.V., & Cunningham-Rather, J. (1984). Complications, consent and cognitions in sex between children and adults. International Journal of Law and Psychiatry,7, 189-203.
3. Auburn, T., & Lea, S. (2003). Doing cognitive distortions: A discursive psychology analysis of sex offender treatment talk. British Journal of Social Psychology, 42, 281-298.
4. Beail, N. (1985). An introduction to repertory grid technique. In N. Beail (Ed.), Repertory Grid Technique and Personal Constructs: Applications in Clinical and Educational Settings. (pgs. 1-24). Cambridge: Brookline Books.
5. Becker, J.S. (2011). Cognitive behavior therapy: Basics and beyond. New York: Guilford Press.
6. Bell, R.C. (2005). The repertory grid technique. In F. Fransella (Ed.), The Essential Practitioner's Handbook of Personal Construct Psychology (pgs. 67-76. Chichester: Wiley.
7. Blagden, N., Winder, B., Gregson, M., & Thorne, K. (2012). The practical utility of using repertory grids with sexual offenders maintain their innocence: A case study. The British Journal of Forensic Practice, 14, 269-281.
8. Blumenthal, N., Gudjonsson, G., & Burns, J. (1999). Cognitive distortions and blame attribution in sex offenders against adults and children. Child Abuse and Neglect, 23, 129-143.
9. Broxholme, S.L. & Lindsay, W.R. (2003) Development and preliminary evaluation of a questionnaire on cognitions related to sex offending for use with individuals who have mild intellectual disabilities. Journal of Intellectual Disability Research, 47, 472-482.
10. Bumby, K. (1996). Assessing the cognitive distortions of child molesters and rapists. Development and validation of the RAPE and MOLEST scales. Sexual Abuse: A Journal of Research and Treatment, 8, 37-54.
11. Daniel, W.W. (1999). Biostatistics: A foundation for analysis in the health sciences. New York: John Wiley & Sons.
12. Doren, D.M. & Yates, P.M (2008). Effectiveness of sex offender treatment for psychopathic sexual offenders. International Journal of Offender Therapy and Comparative Criminality, 52, 234-245.
13. Egan, V., Kavanagh, B., & Blair, M. (2005). Sexual offenders against children: The influence of personality and obsessionality on cognitive distortions. Sexual Abuse: A Journal of Research and Treatment, 17, 223-240.
14. Finkelhor, D. (1984). Child sexual abuse: New theory and research. New York: Free Press.
15. Gannon, T.A., & Alleyne, E.K.A. (2013). Female sexual abusers' cognition: A systematic review. Trauma, Violence, & Abuse, 14, 67-79.
16. Gannon, T.A., Keown, K., & Rose, M.R. (2009). An examination of current psychometric assessments of child molester offense-supportive beliefs using Ward's implicit theories. International Journal of Offender Therapy and Comparative Criminology, 53, 316-333.
17. Gannon, T.A., & Polaschek, D.L.L. (2005). Do child molesters deliberately fake good on cognitive distortion questionnaires? An information processing-based investigation. Sexual Abuse: A Journal of Research and Treatment, 17, 183-200.
18. Grice, J.W. (2002). Idiogrid: software for the management and analysis of repertory grids. Behavior Research Methods, 34, 338-341.
19. Grice, J.W., & Assad, K.K. (2009). Generalized Procrustes Analysis: A tool for exploring aggregates and persons. Applied Multivariate Research, 13, 93-112.
20. Hanson, R.K., Bourgon, G., Helmus, L., & Hodgson, S. (2009). The principles of effective correctional treatment also apply to sexual offenders: A meta-analysis. Criminal Justice and Behavior, 36, 865-891.
21. Hanson, R.K., & Morton-Bourgon, K.E. (2004). Predictors of sexual recidivism: An updated meta-analysis (Corrections Research User Report No. 2004-02). Ottawa, ON: Public Safety and Emergency Preparedness Canada.
22. Hanson, R.K., Gordon, A., Harris, A.J.R., Marques, J.K., Murphy, W., Quinsey, V.L., & Seto, M.C. (2002). First report of the Collaborative Data Outcome Project on the effectiveness of psychological assessment of for sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 14, 169-194.
23. Hanson, R.K., Gizzarelli, R., & Scott, H. (1994). The attitudes of incest offenders: Sexual entitlement and acceptance of sex with children. Criminal Justice and Behavior, 21, 187-202.
24. Hare, R.D. (2003). Hare psychopathy checklist-revised: 2nd edition. North Tonawanda, New York: Multi-Health Systems, Inc.
25. Harkins, L., Beech, A.R., & Thornton, D. (2013). The influence of risk and psychopathy on the therapeutic climate in sex offender treatment. Sexual Abuse: A Journal of Research and Treatment, 25, 103-122.
26. Helmus, L., Hanson, R.K., Babchisin, K.M., & Mann, R.E. (2013). Attitudes supportive of sexual offending predict recidivism: A meta-analysis. Trauma Violence Abuse, 14, 34-53.
27. Hermann, C.A., Babchishin, K. M., Nunes, K.L., Leth-Steensen, & Cortoni, F. (2012). Factor structure of the Bumby Rape Scale: A two-factor model. Criminal Justice and Behavior,39, 869-886.
28. Hildebrand, M., de Ruiter, C., & de Vogel, V. (2004). Psychopathy and sexual deviance in treated rapists: Association with sexual and nonsexual recidivism. Sexual Abuse: A Journal of Research and Treatment, 16, 1-24.
29. Houston, J. (1998). Making sense with offenders. Chichester: John Wiley.
30. Jankowicz, D. (2004). The easy guide to repertory grids. Chichester: John Wiley.
31. Jennings, J.L., & Deming, A. (2013). Effectively utilizing the "Behavioral" in cognitive-behavioral group therapy of sex offenders. International Journal of Behavioral Consultation and Therapy, 8, 7-13.
32. Kelly, G.A. (2003). A brief introduction to personal construct theory. In F. Fransella (Ed.), International Handbook of Personal Construct Psychology (pgs. 3-20). Cichester : John Wiley.
33. Kelly, G. A. (1955/1991). The psychology of personal constructs. New York: W.W. Norton.
34. Lambrick, F. & Glaser, W. (2004). Sex offenders with an intellectual disability. Sexual Abuse: A Journal of research and Treatment, 16, 381-392.
35. Langdon, P.E., & Talbot, T.J. (2006). Locus of control and sex offenders with intellectual disability. International Journal of Offender Therapy and Comparative Criminology, 50, 391-402.
36. Lösel, F., & Schmucker, M. (2005). The effectiveness of treatment for sexual offenders: A comprehensive meta-analysis. Journal of ExperimentalCriminology,1, 117-146.
37. Mann, R., Webster, S., Wakeling, H., & Marshall, W.L. (2007). The measurement and influence of child sexual abuse supportive beliefs. Psychology, Crime & Law, 13(5), 443-458.
38. Mann, R., & Webster, S. (2012, October). How paedophiles construe their sexual interests. Paper presented at the 31st Annual Conference of the Association for the Treatment of Sexual Abusers (ATSA), Denver, Colorado, United States.
39. Marshall, W.L., Marshall, L.E., Serran, G.A., & O'Brien, M.D. (2011). Rehabilitating sexual offenders: A strength-based approach. Washington D.C.: American Psychological Association.
40. Marshall, W.L., Anderson, D. , & Fernandez, Y.M (1999). Cognitive-behavioral treatment of sexual offenders. Cichester: John Wiley.
41. Milhailides, S., Devilly, G.J., & Ward, T. (2004). Implicit cognitive distortions and sexual offending. Sexual Abuse: A Journal of Research and Treatment,16, 333-350.
42. Murphy, W.D. (1990). Assessment and modification of cognitive distortions in sex offenders. In W.L. Marshall, D.R. Laws, & H.E. Barbaree (Eds.), Handbook of Sexual Assault (pp. 331-342) New York: Plenum.
43. Neimeyer, R.A. (2004). Repertory grid methods. Available at www.pcp-net.org/encyclopaedia/repgrid-methods.html.
44. Rice, M.E., & Harris, G.T. (1997). Cross-validation and extension of the Violence Risk Appraisal Guide for child molesters and rapists. Law and Human Behavior, 21, 231-241.
45. Roese, N.J., & Vohs, K.D. (2012). Hindsight bias. Perspective on Psychological Service, 7, 411-426.
46. Strickland, S. (2008). Female sex offenders: Exploring issues of personality, trauma, and cognitive distortions. Journal of Interpersonal Violence, 23, 474-489.
47. Walker, B.M., & Winter, D.A. (2007). The elaboration of personal construct psychology. Annual Review of Psychology, 58, 453-477.
48. Walker, L.J. (1980). Cognitive and perspective-taking prerequisites for moral development. Child Development, 51, 131-139.
49. Ward, T. (2000). Sexual offenders' cognitive distortions as implicit theories. Aggression and Violent Behavior, 5, 491-507.
50. Ward, T., & Keenan, T. (1999). Child molesters' implicit theories. Journal of Interpersonal Violence, 14,821-838.
51. Ward, T., Hudson, S.M., Johnston, L., & Marshall, W.L. (1997). Cognitive distortions in sex offenders: An integrative review. Clinical Psychology Review, 17, 479-507.

Author address

Dawn Pflugradt
Sand Ridge Secure Treatment Center
1111 North Road; Mauston, WI 53948
(608) 847-4438 ext 2136
Dawn.pflugradt@dhs.wisconsin.gov